We used R version 4.0.3 (2020-10-10) and the R-packages tidyverse (version 1.3.0), here (version 0.1), brms (version 2.14.4), modelr (version0.1.8), tidybayes (version 2.3.1), and patchwork (version 1.0.1) for data preparation and analysis.
Five hundred and seventy-one participants were recruited to take part in this study online via Qualtrics. Of which 344 provided full informed consent. One hundred and fifty participants were excluded from this sample due to having completed less than 90% of the questionnaire, providing invalid employment details (i.e. stating they were both employed and unemployed) or for reporting having played no games before or during lockdown. A further 39 participants were removed from the analysis due to having more than 20% of trials with missing data and/or having reported hours played more than 3 MAD above the median hours played in games in an average week (i.e. around 150 hours). After all exclusions we analysed data from 155 participants (age M = 32.581, SD = 8.938, Range = 19 - 72). On average participants took 23.842 minutes to complete the task (SD = 58.43).
The below graph shows the number of participants in a given employment situation during lockdown.
Count of participants for reported employment situation.
The below graph shows the number of participants affected by differing lockdown situations.
Count of participants for reported lockdown situation.
The below graph shows the number of participants affected by in a given living situation.
Count of participants for each reported living situation.
We took the data from the DASS questionnaire and loneliness questionnaire ratings before and after lockdown and combined these with hours played before and after lockdown. Given the data are generated from three Likert-style questionnaire responses per subscale, added together and multiplied by two, responses are thus strictly positive integers. This required fitting the data to cumulative models using a logit link function.
We fitted these models separately for each sub-scale of the DASS and for loneliness using the brm function in brms, estimating the effect of hours played in video games, time (pre- and post-lockdown), and the interaction between them. The categorical fixed effect of time was sum-coded (before = -1, after = 1) while the continuous fixed effect of total hours played was z-transformed. As a result, the intercept represents the grand mean and regression coefficients represent the impact of lockdown on mental health outcomes across the average hours played (i.e. a main effect of time), the impact of hours played across the average of both time points (i.e. a main effect of hours played), and their interaction. Similar
All models contained random intercepts per participant. Models used a \(Normal(0, 50)\) prior on the intercept, a \(Normal(0, 5)\) prior on the slope terms, and an \(Exponential(1)\) prior on the standard deviation term. We evaluate the evidence in support for an effect for each fixed factor and their interaction using Bayes factors calculated using the Savage Savage-Dickey density ratio with the hypothesis function in brms. The fixed effect parameter estimates are reported in the Appendix on the log scale and backtransformed to the natural (i.e. rating) scale.
| Model | Hypothesis | Estimate | Est.Error | CI.Lower | CI.Upper | Evid.Ratio | Post.Prob | Star |
|---|---|---|---|---|---|---|---|---|
| Depression | Time | 0.48 | 0.14 | 0.22 | 0.75 | 0.10 | 0.09 | * |
| Depression | Total Hours | 0.33 | 0.22 | -0.10 | 0.75 | 7.20 | 0.88 | NA |
| Depression | Time by Hours | -0.06 | 0.14 | -0.34 | 0.20 | 34.01 | 0.97 | NA |
| Anxiety | Time | 0.18 | 0.14 | -0.09 | 0.44 | 17.10 | 0.94 | NA |
| Anxiety | Total Hours | 0.37 | 0.22 | -0.07 | 0.80 | 5.48 | 0.85 | NA |
| Anxiety | Time by Hours | -0.08 | 0.13 | -0.35 | 0.18 | 32.12 | 0.97 | NA |
| Stress | Time | 0.50 | 0.13 | 0.24 | 0.76 | 0.00 | 0.00 | * |
| Stress | Total Hours | -0.07 | 0.21 | -0.50 | 0.37 | 21.89 | 0.96 | NA |
| Stress | Time by Hours | 0.11 | 0.13 | -0.14 | 0.37 | 26.80 | 0.96 | NA |
| Loneliness | Time | 0.70 | 0.15 | 0.42 | 1.02 | 0.00 | 0.00 | * |
| Loneliness | Total Hours | -0.07 | 0.22 | -0.52 | 0.36 | 22.96 | 0.96 | NA |
| Loneliness | Time by Hours | 0.04 | 0.15 | -0.24 | 0.33 | 33.41 | 0.97 | NA |
We found evidence in support of the null for all effects expect that of a main effect of Time in the Depression model (BF10= 9.883). In this instance, the parameter estimate is negative \(\hat{\beta}\) = 0.48, 95% CI = 0.221 - 0.753, supporting the notion that depression scores decreased as time spent gaming increased.
CHECK THIS!!!!!
We next explored the effect of the change in total hours playing games before and after lockdown on the difference in mental health outcomes pre- and post-lockdown. Here, hours played after were subtracted from hours played before, and DAS outcomes after were (separately) subtracted from DAS outcomes before.
Models were again fitted separately for each subscale in brms using the brm function. Here, the data were fitted using a Gaussian model (identity link function), with the fixed effect of difference in hours played. Models used a \(Normal(0, 5)\) prior on the intercept, a \(Normal(0, 1)\) prior on the slope term, and an \(Exponential(1)\) prior on the sigma term.
Again, the presence of an effect of difference in hours played was evaluated using Bayes factors calculated using the Savage-Dickey density ratio. We also present the fixed effects for these models on the natural scale.
We present posterior predictions for the effect of difference in hours played pre- and post-lockdown on differences in outcomes for each subscale. Here, lines represent the posterior median along with 50%, 80%, and 95% credible intervals (shaded).
CHECK THIS SUMMARY
| Model | Hypothesis | Estimate | Est.Error | CI.Lower | CI.Upper | Evid.Ratio | Post.Prob | Star |
|---|---|---|---|---|---|---|---|---|
| Depression | Hours Before - Hours After | 0.02 | 0.05 | -0.07 | 0.12 | 16.98 | 0.94 | NA |
| Anxiety | Hours Before - Hours After | 0.01 | 0.03 | -0.06 | 0.07 | 29.32 | 0.97 | NA |
| Stress | Hours Before - Hours After | -0.02 | 0.04 | -0.11 | 0.06 | 20.62 | 0.95 | NA |
We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of difference in hours played pre- and post-lockdown and the difference in stress scores.
| Estimate | Est.Error | l-95% CI | u-95% CI | Rhat | Bulk_ESS | Tail_ESS | |
|---|---|---|---|---|---|---|---|
| Intercept | -3.14 | 1.51 | -6.028 | -0.148 | 1 | 3771 | 2611 |
| total_hours_played_after | -0.01 | 0.04 | -0.086 | 0.067 | 1 | 4204 | 2940 |
## # A tibble: 1 x 8
## Hypothesis Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <lgl>
## 1 (total_hours_… -0.0103 0.0401 -0.0857 0.0674 24.5 0.961 NA
We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of total hours played during lockdown and the difference in depression scores.
| Estimate | Est.Error | l-95% CI | u-95% CI | Rhat | Bulk_ESS | Tail_ESS | |
|---|---|---|---|---|---|---|---|
| Intercept | -1.555 | 0.952 | -3.430 | 0.283 | 1 | 3944 | 3166 |
| total_hours_played_after | 0.005 | 0.026 | -0.045 | 0.056 | 1 | 3788 | 2715 |
## # A tibble: 1 x 8
## Hypothesis Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <lgl>
## 1 (total_hours_… 0.00525 0.0259 -0.0450 0.0563 36.8 0.974 NA
We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of total hours played during lockdown and the difference in anxiety scores.
| Estimate | Est.Error | l-95% CI | u-95% CI | Rhat | Bulk_ESS | Tail_ESS | |
|---|---|---|---|---|---|---|---|
| Intercept | -2.235 | 1.273 | -4.659 | 0.293 | 1 | 3501 | 2909 |
| total_hours_played_after | -0.017 | 0.034 | -0.085 | 0.049 | 1 | 3307 | 2986 |
## # A tibble: 1 x 8
## Hypothesis Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <lgl>
## 1 (total_hours_… -0.0174 0.0337 -0.0847 0.0494 26.0 0.963 NA
We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of total hours played during lockdown and the difference in stress scores.
Posterior predictive checks were carried out for all fitted models. These show a relatively good fit for each model whereby draws from the posterior are close to the fitted data. This indicates good model fit.
Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Depression
Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression
Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Anxiety
Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression
Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Stress
Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression
conent to come.
conent to come.